Clustering Technique Based On The Browsing Behavior Of Swarm Intelligence And Realization

The design and management of websites and systems are becoming more and more significant with the rapid improvement of Internet. Useful knowledge extracted from Web logs can be used to improve the structure design of websites, analyze system performance, understand users' behaviors and motivations, and construct personalized websites. Users are the most important resources, website operators should pay much more attention to users' browsing behaviours, and the feature of users' behaviors appears to be particularly critical. Cluster analysis technique, known as one of the main functions of Data Mining and Knowledge Discovery, categorizes data area sets into natural classes and gives a description of features for each class. Through studying users' browsing behaviors characteristics, the cluster analysis method divides users into different classes. Based on careful analysis of the common behaviours for each class, website operators can get a better understanding of their users' behaviors, explore users' potential demands and interests, find the behavioral regularities, hereby improve the structural design of website and provide the distinctive personalized services for users in E-commerce environment.Swarm intelligence is defined as "any attempt to design algorithms or distributed problem-solving devices inspired by the collective behavior of the social insert colonies and other animal societies". The clustering algorithm based on swarm intelligence can achieve better clustering result when comparing with traditional clustering algorithm. The Ant Colony Clustering Algorithm is a novel optimization method and has very strong robustness and adaptiveness. It shows excellent performance and great development potential of evolution in solving complicated optimization problems, especially discrete optimization problems.The dissertation is based on swarm intelligence, integrated with the advanced research achievements from home and abroad, we choose the Optimized Ant Cluster Algorithm which is proposed by Dai Weihui, Liu Shouji et al from Fudan University. We use the web log data collected from internet users' browsing behaviours, pre-process the log data through the following operations: purify data, identify users, identify sessions, complete paths, and so forth. And then perform the pattern recognition and analysis. We do the cluster analysis on users' behaviours. The purpose of the cluster is to identify user base with similar features, and to be the base of user analysis and business strategy. As well as that, we can also provide users with more high-quality services to meet their demands.The dissertation is divided into six chapters. The first one is about research background, research status, research content, and structure of this dissertation; The second chapter is the literature review; The third chapter introduce ant cluster algorithm on swarm intelligence; The fourth part is about users' browsing behaviors and the cluster method used in analyzing it; And then the fifth chapter gives the practical application in network users' behaviors analysis using the Optimized Ant Cluster Algorithm; The last chapter is about the conclusion and future.